WSINDy4Dispersal

Weak form model discovery for insect dispersal.

---

Python code accompanying the manuscript "Weak-Form Learning for Mean-Field Partial Differential Equations: an Application to Insect Movement" submitted to the SIAM Journal on Life Sciences

• Zenodo: https://zenodo.org/records/17156064
• ArXiV pre-print: https://doi.org/10.48550/arXiv.2510.07786

@misc{minor2025weakformlearningmeanfield,
      title={Weak Form Learning for Mean-Field Partial Differential Equations: an Application to Insect Movement}, 
      author={Seth Minor and Bret D. Elderd and Benjamin Van Allen and David M. Bortz and Vanja Dukic},
      year={2025},
      eprint={2510.07786},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2510.07786}, 
}

Open data: see the caterpillars.csv file.

• See the tutorials and examples located here for instructions on how to use the codebase.
• To recreate results in the paper, see the wsindy_for_dispersal.ipynb notebook.

---

Our code uses the following dependencies:

# Core
import torch
import scipy
import numpy as np
import matplotlib
import itertools
import symengine as sp
import pandas as pd

# Special
import torch.linalg as la
from scipy.signal import convolve
from scipy.special import factorial
import matplotlib.pyplot as plt
from tqdm import tqdm
from IPython.display import display,Math

# Weak SINDy
from wsindy import *
from helper_fcns import *

Install PyWSINDy for PDEs a Bash environment:

wget -q https://raw.githubusercontent.com/SethMinor/PyWSINDy-for-PDEs/main/wsindy.py
wget -q https://raw.githubusercontent.com/SethMinor/PyWSINDy-for-PDEs/main/helper_fcns.py